Subversion Repositories SmartDukaan

'''

Created on 20-May-2010

@author: gaurav

'''

from scrapy.spider import BaseSpider

from scrapy.selector import HtmlXPathSelector

from scrapy.http import Request

from demo.items import DemoItem

from scrapy.contrib.spidermiddleware import referer

from scrapy.http.headers import Headers

from scrapy.http.request.form import FormRequest

from scrapy.log import msg

from scrapy.http.response import Response

from datastore import DataAccessor

from datastore.DataAccessor import DataHelper

from html2text.unescaping import *

class mobstore_price(BaseSpider):

    def __init__(self):

       MOBILESTORE_DOMAINNAME1 = "mobilestore1"   

       self.domain_name = MOBILESTORE_DOMAINNAME1 

       # get urls from the database and append them in the list for crawling

       da = DataHelper()

       for pitem in da.get_allmobstoreurls():

            self.start_urls.append(pitem.url.strip())

    def start_requests(self):

        listreq = []

        #for each request a referer has to be set

        MOBILESTORE_REFERER = "www.google.com/search"

        for url1 in self.start_urls:

            request = Request(url = str(url1), callback=self.parse)

            request.headers.setdefault("Referer", MOBILESTORE_REFERER)

            listreq.append(request)

        return listreq

    def parse(self, response):

        site = response.url

        site = unescape(site)

        MOBILESTORE_VATPLUSTAX = 0

        #retreiving model-name from the url

        pos1 = pos2 = 0

        temp = ""

        pos1 = site.rfind('/')

        if pos1 != -1:

            temp = site[pos1+1:len(site)]

        pos3 = temp.find('.')

        temp1 = temp[pos3:len(temp)]

        name = temp.replace(temp1,"")         

        hxs = HtmlXPathSelector(response)

        MOBILESTORE_XPATH2 = '//div[@id ="priceComp"]//tr[2]/td[3]/span/text()' 

        prices = hxs.select(MOBILESTORE_XPATH2)

        #removelist is used for converting price to decimal format containing only numbers and '.'

        MOBILESTORE_REMOVELIST = ["Rs",",","-","/","Rs."]

        da = DataHelper()

        for price in prices:

             name = str(name).strip()

             price = price.extract()

             price = str(price).strip()

             if price != '':        

                for r in MOBILESTORE_REMOVELIST: 

                    while price.find(r) != -1:

                        price = price.replace(r, "")

             price = price.strip()

             shown_pr = int(price)

             final_pr = shown_pr + MOBILESTORE_VATPLUSTAX 

             da.add_new_mobstorephone(name,shown_pr,final_pr)

SPIDER = mobstore_price()